Transparent assessment of information quality of online reviews using formal argumentation theory

Review scores collect users’ opinions in a simple and intuitive manner. However, review are also easily manipulable, hence they often accompanied by explanations. A substantial amount of research has been devoted to ascertaining the quality reviews, identify most useful authentic through explanation analysis. In this paper, we advance state art We introduce rating system arguments define an appropriate weighted semantics formal argumentation theory. algorithm construct corresponding graph, based on selection arguments, their semantic distance, supported ratings. provide model such maximizing overall weight admitted nodes edges. evaluate these contributions Amazon dataset McAuley et al. (2015), comparing results our assessment with upvotes received reviews. Also, deepen evaluation crowdsourcing multidimensional reviews it assessment. Lastly, perform user study explainability method, i.e., test whether automated method use assess is understandable humans. Our achieves two goals: (1) identifies that considered useful, comprehensible, complete online users, does so unsupervised manner, (2) provides assessments.